Providers

This guide explains how to deploy and use the Cassandra Operator on various providers.

Cassandra Operator on Azure Kubernetes Service

This guide explains how to deploy and use the Cassandra Operator on Azure Kubernetes Service (AKS).

Requirements

• An AKS cluster with at least 3 nodes.

When deploying the cluster on AKS, it's important to properly set fsGroup option:

cat <<EOF >cluster.yaml
apiVersion: cassandraoperator.instaclustr.com/v1alpha1
kind: CassandraDataCenter
metadata:
  name: cassandra-test
...
spec:
  fsGroup: 999
EOF

kubectl apply -f cluster.yaml

NOTE: Setting fsGroup to 999 is necessary on AKS. Without it, Azure Disk volumes attached to the Cassandra pods won't be writeable by the Cassandra process.

Example of deployment

Firstly, you need to create a resource group under which all resources will live:

$ az group create --name myResourceGroup --location eastus

Then we have to create AKS cluster, here with two nodes:

$ az aks create \
    --resource-group myResourceGroup 
    --name myAKSCluster \
    --node-count 2 \ 
    --generate-ssh-keys

The creation of a cluster takes a while. In order to connect to such cluster, you need to get credentials so your kubectl will work:

$ az aks get-credentials \
    --resource-group myResourceGroup \
    --name myAKSCluster

After you apply cluster.yaml above, you should have three nodes running:

$ kubectl get pods
NAME                                  READY   STATUS    RESTARTS   AGE
cassandra-cassandra-test-rack1-0      2/2     Running   6          79m
cassandra-cassandra-test-rack1-1      2/2     Running   0          75m
cassandra-cassandra-test-rack1-2      2/2     Running   0          73m
cassandra-operator-78c59b86cd-94jdq   1/1     Running   0          79m

If you want to expose a node, you need to do it like this:

$ kubectl expose pod cassandra-cassandra-test-rack1-0 \
    --type="LoadBalancer" \
    --name=node1-service \
    --port=9042 \
    --target-port=9042

After a while, you see that public IP was assigned, e.g. like this:

kubectl get services
NAME                TYPE           CLUSTER-IP     EXTERNAL-IP      PORT(S)
... other services
node1-service       LoadBalancer   10.0.166.108   52.226.147.217   9042:30462/TCP

So you connect to that node like:

$ cqlsh 52.226.147.217
Connected to cassandra-test at 52.226.147.217:9042.
[cqlsh 5.0.1 | Cassandra 3.11.4 | CQL spec 3.4.4 | Native protocol v4]
Use HELP for help.
cqlsh> 

Please keep in mind that this connection is not secured as it is not running on SSL. You might change authenticator via configuration mechanism beforehand to modify cassandra.yaml so you might create a user with strong password and possibly change password for default cassandra user.

Minikube

1. In the project root directory e.g. cd ~/git/cassandra-operator/ Start a minikube cluster with enough resources, Cassandra is hungry!

   $ minikube start --cpus 4 --memory 4096 --kubernetes-version v1.14
   

2. Use the minikube docker context

   $ eval $(minikube docker-env)
   

   Alternatively, use other VM/kvm based installs to create a local k8 environment.

3. If multi-rack is wanted, add labels to the local "nodes" to allow multi-rack pod placement. Have a look at Custom configurations for configuration options for the Cassandra clusters.

4. Add Persistent Volumes to your environment. One way to do so would be using NFS-mounted volumes for the k8 environment via sharing local folders. For example, by creating and exposing /volumes/00 to the kubernetes network, one could use the following yaml fragment to create the PV:

   apiVersion: v1
   kind: PersistentVolume
   metadata:
     name: nfs00
   spec:
     capacity:
       storage: 1Gi
     accessModes:
       - ReadWriteOnce
       - ReadWriteMany
     persistentVolumeReclaimPolicy: Recycle
     nfs:
       server: <localhost IP in k8 network>
       path: "/volumes/00"

   You can repeat this action as many times as required for many folders, creating multiple persistent volumes.